Segmentation fault while running dmr-hdhmm

[Ian Chang]

Hi all, 

When I run dmr-hdhmm, I got a segmentation fault. And the error occurs even with smaller samples (200000 cpgs). 

However the methcount files works find with hmr program.

Do you have any idea or suggestion to solve the problem?

Thank you so much.

Ian

$dmr-hdhmm  -A mm9_A.meth -B mm9_B.meth -o test.out -s test.score -v

Reading input file mm9_A.meth ...  Done

Reading input file mm9_B.meth ...  Done

Calculating diffscores: Probes retained: 0

[SEPARATING BY CPG DESERT]

CPGS RETAINED: 0

DESERTS REMOVED: 18446744073709551615

Segmentation fault

[Song, Qiang]

Hi Ian,

It seems that there is no CpG sites common to both input files that are covered in both files.
It may happen when input file A and input file B come from different regions, or they all have very low coverage.
What is the output when you compare those two files with methdiff?

Best,
Song Qiang

--
You received this message because you are subscribed to the Google Groups "MethPipe and MethBase Users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to methpipe+u...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[Ian Chang]

Dear Song,

The CpG sites are all identical in two methcount files.

The output of methdiff looks fine.

I attached 200,000 CpG sites of two methcount files in the post. Could you please test them if you have time.

The methpipe version is 3.3.1.

GCC version is 4.7.2 

Linux Host 2.6.32.12-0.7-default #1 SMP 2010-05-20 11:14:20 +0200 x86_64 GNU/Linux

Thank you so much.

Ian

[Ian Chang]

Dear Song, 

I found that the problem may due to get_meth_unmeth function.

    const string name(cpg.get_name()); --> will return "CpG" only

    const size_t n_reads = atoi(name.substr(name.find_first_of(":") + 1).c_str()); --> will return zero

Then I traced the problem and found that in GenomicRegion.cpp line 252.

    original: name = parts[3];

    revised: name = parts[3]+":"+parts[5];

I think this should be the bug.

 

Ian